Validation of the adolescent Reading Motivation and Preferences Scale

Design

A non-probability, intentional sampling method was used for this research. On the one hand, in non-probability sampling, not all elements of the target population have the same probability of being selected. Unlike probability sampling, in this case the selection of elements is not based on random methods, which means that the results may not be representative of the entire population. On the other hand, purposive sampling is a specific type of non-probability sampling in which the elements considered most suitable for the study are deliberately selected. The sample selected (835 participants) corresponds to a 95% confidence interval and a 3.4% margin of error.

Participants

The sample population consists of a total of 835 subjects from 25 secondary schools on the island of Gran Canaria (Las Palmas, Spain). Among them, 43.8% were boys and 53.4% were girls, while 2.8% of participants identified as non-binary, meaning that their gender identity does not fit into the gender binary (male or female). The average age of participants is 15.1 (ranging from 12 to 19 years old) and they are enrolled in secondary education (54.3%) and upper secondary education (44.3%).

Procedure

Participation in the study was made possible thanks to the collaboration of the management teams and teaching staff at the participating schools, who authorised the collection of data and supervised the completion of the questionnaire during school hours. Informed consent was obtained: concise information about the purpose and academic aim of the study, ethical protocol, and guarantee of anonymity and confidentiality throughout the process (Creswell, 2007).

From the outset of the research, the operational structure of the educational establishments was taken into account and efforts were made to facilitate the collection of data from students. To this end, an online form was created using Google Forms. In this way, students, accompanied by teachers, could access the programme via computers in the computer room, tablets loaned by the school, or their own devices. At the same time, the competence level of the informants was taken into account, so that during the development of the scale, special care was taken in the wording of the items to ensure their comprehension by adolescents at different educational levels.

Instrument

The EMyPLA Scale is an ad hoc instrument designed to explore the motivations that drive reading for pleasure in adolescence, as well as reading preferences associated with genres and reading formats. Its final design integrated the prescribed sociodemographic profile section and a specific section (Table 1) that, following intentional criteria, chose a factorial structure of six elements and a total of 31 closed-ended questions, scaled with a six-category Likert scale:

1. Never, 2. Hardly ever, 3. Sometimes, 4. Quite often, 5. Almost always, 6. Always.

Data analysis

Initially, both exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were carried out, using the exploratory structural equation model (ESEM). To facilitate interpretation of the factors, the orthogonal geomin method was chosen (Schmitt, 2011).

According to the fit indices, the model is considered well fitted when the CMIN ( ${\chi }^2$/gl) is less than 2; the RMSEA is less than or equal to .05 or less than or equal to .08 when the upper limit of the confidence interval is not greater than .08; the SRMR is less than .08, the CFI is equal to or greater than .90, and the TLI is equal to or greater than .95 (Kline, 2015). Subsequently, reliability evidence was analysed using McDonald's omega (Zinbarg et al., 2005). In addition, the normality of the data was examined by analysing the asymmetry and kurtosis of each factor.

The relationship between the factors was evaluated using Pearson's correlation analysis, which allowed us to verify whether the different motivations and reading preferences identified in the scale were related to each other. All correlations were statistically significant (p < .01), indicating that, although the factors represent different constructs, there is adequate internal consistency among them. To verify that the identified factors measured distinct elements of motivation and distinct reading preferences of the construct, a repeated measures analysis (intra-subject ANOVA) was conducted, comparing the mean scores obtained by the students on each of the factors.

Subsequently, t-tests were carried out on paired samples of factors in order to identify which combinations showed significant differences. Results with p < .05 were considered significant, and effect sizes were reported using the eta squared statistic ( ${\eta }^2$) (Lenhard & Lenhard, 2016).

Validation of the Scale of Reading Motivation and Preferences in Adolescents

In this section, we first analysed the Kaiser-Meyer-Olkin (KMO) values and Bartlett's sphericity test. The KMO yielded an acceptable value of 0.762, while the Bartlett test obtained a value of 1439.903 (p < .001). These results support the feasibility of conducting an exploratory factor analysis.

The normality of the data then was analysed before applying parametric analyses. To this end, the Kolmogorov-Smirnov and Shapiro-Wilk tests were performed, which were statistically significant in all factors (p < .001). Although this could suggest deviations from normal distribution, it should be noted that these tests tend to be overly sensitive in large samples, such as the one in this study (N = 835), detecting trivial deviations that are not relevant from a practical point of view (Ghasemi & Zahediasl, 2012). Therefore, this analysis was complemented with asymmetry and kurtosis values, which remained within acceptable ranges (asymmetry between -2 and +2; kurtosis between -7 and +7), allowing us to assume a reasonably normal distribution of the data: the asymmetry for the 31 items on the scale ranges from -31 to -2,864, while the kurtosis varies between -1.54 and 9.82.31 and 2.864, while kurtosis varies between -1.54 and 9.82. Although one of the items exceeds these limits, it was decided to retain it because of its theoretical importance, given that it addresses the influence of teachers as one of the contextual factors in adolescent reading motivation.

A Pearson correlation analysis was performed between the six factors identified in the scale, with the aim of examining the relationship between the different motivational and preference elements defined by the factor structure. All factors were significantly correlated with each other (p < .001), indicating adequate internal consistency. The highest correlations were observed between Imaginative-creative motivation and Recreational-communicative motivation (r = .731), as well as between Novel genres and both motivational factors (r = .507 and r = .495, respectively).

These results show that, although each factor represents a different dimension of reading in adolescence, there is a significant connection between them, which supports the structural validity of the proposed model.

Next, exploratory and confirmatory factor analyses were performed, evaluating factor structures ranging from 1 to 10 factors. The structure with adequate fit indices was selected, where each factor had at least two significant items; in this case, the correlation of the items with the factor must be equal to or greater than .07. For this scale, the six-factor model (table 1) proved to be the most appropriate factorial solution.

The original questionnaire consisted of 36 items. After factor analysis, five items were discarded because they did not meet the statistical significance criteria required for factor configuration. The following were removed: ‘I usually look up words I don’t understand in the dictionary’, ‘When I like a book and find it entertaining, I find time during the day to read it’, ‘I read about my hobbies to learn more about them’, ‘I like writing stories and tales about the books I read’ and ‘I like learning new things through books’.

These items were excluded from the final model because they had factor loadings below .30 or showed similar loadings on more than one factor, without a minimum difference of .15 between them, which prevented their clear assignment to a single element. The remaining 31 items comprise the final version of the scale presented and analysed in this study.

The factorial structure composed of six factors explains a total of 61.52% of the variance of the instrument, indicating that the model has adequate explanatory power.

In detail, the first factor explains 32.26% of the variance, followed by the second (10.21%), the third (5.43%), the fourth (5.21%), the fifth (4.61%) and the sixth (3.79%). This cumulative percentage is considered satisfactory in exploratory psychometric studies and supports the suitability of the proposed model.

The resulting factor solution consists of a total of 31 items, with factor weights ranging from .42 to .97. Likewise, its adjustment values are considered adequate: RMSEA=.05; CI .05-.06; CFI=.96; TLI=.95 and SRMR=.03. The scale has adequate reliability both as a whole (ω=.96) and in each of its component factors. This ranged between .90 and .71: the first factor, Journalistic genres and reading on digital platforms (ω=.81); the second, Novel genres (ω=.80); the third, Imaginative-creative motivation (ω=.90); the fourth, Recreational-communicative motivation (ω=.85); the fifth, lyric genre and self-help books (ω=.73); and the sixth, books by YouTubers/streamers (ω=.71).

In order to facilitate understanding of the factors identified in the model, a summary table is provided below with their names, descriptions and number of associated items (table 2).

The factorial weights of the 31 items that make up the final version of the scale are presented below (table 3).

The first factor, Journalistic genres and reading on digital platforms, consists of seven items and covers reading various journalistic genres, as well as reading on digital platforms, including blogs, Twitter, and other documents hosted on the Internet. A high score in this factor reflects greater interest among adolescents in accessing information through reading in different media and platforms, while a low score reveals low interest in reading this type of text.

Next, the second factor, Novel genres, consists of four items related to reading different novel genres. A high score on this factor indicates that adolescents have a greater appreciation for fictional narrative genres, while a low score reflects a low interest in these literary genres.

The third factor, imaginative-creative motivation, consists of seven items and considers reading as an activity that can stimulate the reader's creativity, allowing them to come up with alternative narratives or project themselves into the story. This factor is closely related to participation in fan communities (usually online) around a book or series, where members recreate fictional worlds, often of a fantastical nature. A high score in this factor indicates a greater interest among adolescents in using reading as a means to imagine, create and project themselves into the work; a low score reveals less interest in this dimension of reading.

The fourth factor, Recreational-Communicative Motivation, consists of 8 items and focuses on reading for pleasure as a form of entertainment and an opportunity to communicate with others. A high score on this factor suggests that participants view reading as a form of entertainment and an opportunity to share their reading experiences with those around them. A low score indicates less interest in the recreational and social aspects of reading.

The fifth factor, Lyric genre and self-help books, consists of three items linked to both lyric genres, such as poetry and poetic prose, and self-help and personal development books. Both types of reading share the goal of expressing deep feelings and emotions. A high score on this factor indicates a greater appreciation for reflective and meditative reading about oneself and one's emotions, while a low score reveals less interest in this type of introspective reading.

Finally, the sixth factor, Books by YouTubers/streamers, includes two items and refers to interest in books published by influencers with online visibility, such as YouTubers and streamers. A high score on this factor reveals an interest in the posts made by these online content creators and indicates a desire among teenagers to consume posts by YouTubers/streamers; a low score indicates less interest in this type of post.

After completing the factor analysis and understanding the meaning of each factor, it is crucial to validate the scale by calculating descriptive values, such as the central tendency and distribution of the variables that comprise it. It is also necessary to confirm that each factor measures different aspects. This is achieved by analysing repeated measures and t-tests for related samples.

As for the difference between factors, a repeated measures analysis (intra-subject ANOVA) was performed to compare the mean scores obtained by the students in each of the six factors. The results showed statistically significant differences between the factors (F (5, 4170) = 164.40; p < .001; ${\eta }^2$ = .165), with a high effect size, indicating that students showed different patterns of preference or motivation depending on the type of reading motivation or reading preferences associated with the different factors.

Specifically, the factor with the highest average score was Novel genres (M = 3.48), while the factor with the lowest score was Lyric genres and self-help books (M = 2.12). These results do not directly affect the structural validity of the scale, but they do provide relevant information about the reading trends of the students. The average for this factor, Novel genres, is higher than the factor Journalistic genre and reading on digital platforms (t (834) = 15.38; p < .001); Imaginative-creative motivation (t (834) = 6.23; p < .001); Recreational-communicative motivation (t (834) = 15.7; p < .001); Lyric genre and self-help books (t (834) = 26.08; p < .001) and that the factor Books by YouTubers/streamers (t (834) = 15.12; p < .001).

Upon analysing each factor individually, significant differences between all factors except between the factor Journalistic genres and reading on digital platforms and the factor Recreational-communicative motivation were found. Among genres, the Novel genres factor has a higher score than the other genres. At the same time, the Imaginative-creative motivation factor has a higher score than the Recreational-communicative motivation factor.

Table 4 shows the means, standard deviations and distribution measures for each of the factors: